Implicit conversion from float -109.25373134328359 to int loses precision